Can Large Language Models Solve Optimisation Problems?
A new study explores how GPT-3.5 can generate and debug optimisation models in MiniZinc, bringing us closer to automated problem-solving.
Things I'm Exploring
Latest ideas & research
1 min read
Boris Leonardo
AutoMH - When Computer Programs Learn to Write Better Computer Programs
How computer programs learned to create better computer programs all by themselves, like digital evolution happening right before our eyes
1 min read
Boris Leonardo
The Aerial Acrobats: How Hummingbird Flight Patterns Inspire Revolutionary Optimization Algorithms That Solve Complex Problems
A groundbreaking bio-inspired algorithm captures the extraordinary maneuverability of hummingbirdsâtheir translations, rotations, and complex turnsâtransforming these aerial acrobatics into powerful computational tools for solving optimization challenges that stump traditional methods.
1 min read
Boris Leonardo
Digital Puzzle Masters: How Computer Programs Spent Six Days Solving the Ultimate Factory Organization Challenge with Mathematical Perfection
Scientists created systematic experiments that searched through billions of factory arrangements for six days straight to find the mathematically perfect way to organize machines and parts into three manufacturing cells, achieving optimal solutions for the first time.
1 min read
Boris Leonardo
Digital Andean Condors Soaring Through Mathematical Landscapes: How South America's Master Gliders Taught Computers to Navigate Infinite Optimization Spaces
Scientists created computer programs that explore continuous mathematical landscapes using the same soaring strategies Andean condors use to search vast mountain territories, achieving near-perfect solutions with elegant simplicity.
1 min read
Boris Leonardo
The Digital Rangers: How Artificial Intelligence Learned to Protect Virtual Ecosystems from Collapse
A revolutionary breakthrough in artificial intelligence demonstrates how autonomous multi-agent systems can prevent ecosystem collapse by dynamically regulating predator-prey populations, offering new insights for conservation and environmental management in an era of climate change.
Things I'm Building
Active & completed work
ResearchCompleted
LLMs for Optimisation Problems
Can AI language models solve complex optimization problems? Exploring whether GPT and other transformers can generate constraint programming models that tackle real-world challenges.
Learn more
ResearchCompleted
AutoMH - Teaching AI to Design Better Algorithms
What if algorithms could design themselves? AutoMH uses reinforcement learning to automatically create evolutionary metaheuristic algorithms that adapt and improveâsometimes outperforming human-designed solutions.
Learn more
ResearchActive
Evolutionary Algorithms Research
Evolutionary algorithms mimic natural selection to solve complex optimization problems. This ongoing research explores novel operators, hybrid approaches, and applications across diverse domainsâfrom engineering design to machine learning optimization.
Learn more